three essays in corporate finance

Since these legal liabilities are unrelated to currentoperations, …rms that are in …nancial distress due to their legal woes provide a naturalexperiment to study the impact of …nancial d

Three Essays in Corporate Finance

Dissertation

Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy

in the Graduate School of The Ohio State University

By Jérôme P A Taillard, Dipl

Graduate Program in Business Administration

The Ohio State University

2010

Dissertation Committee:

René M Stulz, Advisor Bernadette A Minton Karen H Wruck Ruediger Fahlenbrach

c Copyright byJérôme P A Taillard

2010

In my dissertation, I …rst contribute to the capital structure literature by mating the potential impact of …nancial distress on a …rm’s real business operations.Secondly, I contribute to the ownership structure literature, and more broadly tothe …eld of corporate governance, by revisiting the relationship between managerialownership and …rm performance

esti-In corporate …nance, a common empirical challenge is to disentangle the tionship of interest from other potential confounding e¤ects In particular, omittedvariables can bias classic econometric estimations and, more generally, cast doubt

rela-on any causal inferences being made by researchers These endogeneity crela-oncerns arepotentially severe within the context of my research and one of the main contribu-tions of my work is to devise two novel empirical approaches to circumvent, at leastpartially, the issue of endogeneity

In my …rst essay, I analyze a comprehensive sample of defendant …rms that foundthemselves exposed to an unexpected wave of asbestos litigation in the wake of twoU.S Supreme Court decisions Since these legal liabilities are unrelated to currentoperations, …rms that are in …nancial distress due to their legal woes provide a naturalexperiment to study the impact of …nancial distress on a …rm’s operational perfor-mance When analyzing …rms su¤ering from this exogenous shock to their …nances, I

…nd little evidence of negative spillover e¤ects (“indirect”costs) of …nancial distress

That is, the competitive position of the distressed …rms is not adversely impacted

by their weakened …nancial situation Furthermore, I …nd empirical support for asigni…cant disciplinary e¤ect of …nancial distress as these …rms actively restructureand refocus on core operations

In my second and third essays, I focus on the relationship between managerialownership and …rm performance using a large panel dataset of U.S …rms over theperiod 1988-2004

In the second essay, I reconcile some of the extant literature by showing that therelationship is sensitive to the …rm size characteristics of the sample being used Inparticular, I recover the classic hump-shaped relationship when focusing only on thelargest …rms (e.g Fortune 500 …rms), while the relationship turns negative when thesample is comprised of smaller …rms

The negative relationship among smaller …rms is consistent with entrenchment guments given that managerial ownership is on average much higher for small …rms.Second, I …nd that for lower levels of managerial ownership, the negative relation-ship is driven by older …rms that have on average less liquid stocks This …nding isconsistent with …rms that do not perform well enough to create a liquid market fortheir stock, and hence have to keep high levels of insider ownership in order to avoid

ar-a negar-ative price impar-act thar-at would result from ar-a reduction of their star-ake Lar-astly,these results could also be suggestive of endogeneity concerns I investigate this issuefurther in my third essay

Principal-agent models predict that managerial ownership and …rm performanceare endogenously determined by exogenous changes in a …rm’s contracting environ-ment Changes in the contracting environment are, however, only partially observed,

and the standard statistical techniques used to address endogeneity may be ine¤ective

in this corporate setting In my third essay, together with my coauthor Phil Davies,

we develop a novel econometric approach to control for the in‡uence of time-varyingunobserved variables related to a …rm’s contracting environment Using the samelarge panel dataset of U.S …rms over the period 1988-2004, we …nd no evidence of asystematic relation between managerial ownership and performance

To my parents, Wendy and André

I wish to thank my dissertation committee members Bernadette Minton, KarenWruck, Ruediger Fahlenbrach and, in particular, my advisor Professor René Stulz fortheir guidance and patience, and for sharing their wisdom and knowledge with me Iwant to thank also Michel Dubois for nurturing my initial interest in …nance and forencouraging me to pursue Graduate studies, and also Greg Allenby for sharing hispassion for Bayesian statistics

I want to thank my family for their unconditional love and encouragement out this academic venture My parents, Wendy and André, have always been there for

through-me and their love has given through-me the strength needed to accomplish this dissertation

My brother Serge and my grandmother Iris have also provided unwavering moralsupport over the years I thank them all for helping me …nd the balance necessary tosucceed in the long run

Among friends and colleagues, I would like to thank …rst Mamata, who stood

by my side throughout and has always encouraged me to go forward I’ll always begrateful for having made this journey with her I would also like to thank Chuan Liaoand Taylor Nadauld from my PhD cohort, and Patrick Huguenin and David Ardia

in Switzerland The co-author on my third chapter, Phil Davies, deserves a specialmention I am very thankful for his support and feel blessed to have found such agood friend

There are many other people who have also helped along the way, including AnilMakhija, Robyn Scholl and the entire Finance Faculty and sta¤ at the Fisher College

of Business I cannot cite them all here but I do want to thank them collectively forall their encouragements, time and e¤ort

May 24, 1979 Born — La Chaux-de-Fonds,

Switzer-land

2002 Diploma in Mathematics applied to

Fi-nance — University of Neuchâtel

2004 Certi…cate of Study in Economics —

Gerzensee, Switzerland

FIELDS OF STUDY

Major Field: Business Administration

Concentration: Finance

TABLE OF CONTENTS

Page

Abstract ii

Dedication v

Acknowledgments vi

Vita viii

List of Tables xii

List of Figures xiv

Chapters: 1 Introduction 1

2 Thriving in the midst of …nancial distress? An analysis of …rms exposed to asbestos litigation 5

2.1 Introduction 5

2.2 Asbestos mass tort litigation 11

2.2.1 The asbestos litigation crisis after Amchem (1997) and Ortiz (1999) 12

2.3 Data 14

2.4 Financial impact of litigation on defendant …rms 16

2.4.1 Measuring the …nancial impact of litigation beyond insurance coverage 16

2.4.2 Litigation exposure and …nancial distress 19

2.5 Analysis of …nancially distressed …rms due to asbestos litigation 24

2.5.2 Timeline in distress 27

2.5.3 Direct costs of litigation and …nancial distress 29

2.5.4 Value computations 30

2.5.5 External …nancing during crisis 34

2.5.6 Quantitative evidence of bene…ts and costs of distress at the operational level 36

2.5.7 Qualitative evidence of bene…ts and costs of distress at the operational level 39

2.6 Conclusion 42

3 Managerial Ownership and Firm Performance Revisited: Taking Cues from Firm Size Heterogeneity 77

3.1 Introduction 77

3.2 Data 81

3.2.1 Construction of the panel dataset 82

3.2.2 Variable de…nition 83

3.2.3 Sample and variable characteristics 85

3.3 Methodology 89

3.3.1 Baseline speci…cation 89

3.3.2 Non-linearities 92

3.3.3 Instrumental variable (IV) estimation 94

3.4 Results 97

3.4.1 Baseline OLS speci…cation 98

3.4.2 Instrumental variable (IV) speci…cation 101

3.4.3 Time-series evidence 103

3.4.4 Analyzing the relationship across …rm size 105

3.4.5 Understanding the changes in the relationship related to …rm size 109

3.5 Conclusions 111

4 Omitted Variables, Endogeneity, and the Link Between Managerial Own-ership and Firm Performance 137

4.1 Introduction 137

4.2 Literature review 142

4.3 Methodology 145

4.3.1 Model Overview 145

4.3.2 Simulation Study 148

4.4 Data 151

4.4.1 Construction of the panel dataset 151

4.4.2 Variable de…nition 152

4.4.3 Sample and variable characteristics 154

4.5 Results 157

4.5.1 Full sample 158

4.5.2 Heterogeneity 163

4.6 Conclusions 168

Bibliography 187

Appendices: A Additional Material for Chapter 2 195

A.1 Data description 196

A.2 Instrumental variable (IV) approach 198

B Additional Material for Chapter 3 202

B.1 Data description 203

B.2 Estimation Algorithm 207

LIST OF TABLES

2.1 Defendant …rm and litigation characteristics 49

2.2 Distribution of litigation exposure among defendant …rms 52

2.3 Detecting …nancial distress due to asbestos litigation 54

2.4 Sample of …rms su¤ering from …nancial distress due to litigation 57

2.5 Characteristic comparison between distressed and non-distressed de-fendants 60

2.6 Litigation costs, direct and estimated net indirect costs of …nancial distress 62

2.7 Shareholders’returns from pre-distress to post-resolution 64

2.8 Evidence on …nancing 66

2.9 Operating performance of distressed …rms 69

2.10 Measures of operational e¢ ciency 71

2.11 Evolution of discretionary expenditures 73

2.12 Qualitative evidence of the bene…ts and costs of …nancial distress 76

3.1 Descriptive statistics 120

3.2 Descriptive statistics by …rm size 122

3.3 Ordinary Least Squares (OLS) estimations 124

3.4 Instrumental variable (IV) estimations 127

3.5 Yearly regressions 130

3.6 Size regressions 132

3.7 OLS comparisons across di¤erent …rm size categorizations 134

3.8 Time since IPO within smallest …rms category 136

4.1 Simulation study 171

4.2 Descriptive statistics 173

4.3 Main speci…cation 175

4.4 Main speci…cation with the logarithm of managerial ownership 177

4.5 Robustness checks on the baseline model 179

4.6 Model estimation by industry 181

4.7 Model estimation by …rm size 183

4.8 Results for controls by …rm size 185

LIST OF FIGURES

2.1 Evolution of asbestos litigation activity 45

2.2 Litigation strategy for defendant …rms 46

2.3 Evolution of net out‡ows due to litigation 47

3.1 Distribution of managerial ownership 114

3.2 Managerial ownership distribution across size categories 115

3.3 Evolution of managerial ownership across size categories 116

3.4 Firm size distribution for each category given by the Morck et al cuto¤s117 3.5 Stock liquidity for small …rms across young and old cohorts 118

CHAPTER 1

INTRODUCTION

In my dissertation, I …rst contribute to the capital structure literature by mating the potential impact of …nancial distress on a …rm’s real business operations.Secondly, I contribute to the ownership structure literature, and more broadly tothe …eld of corporate governance, by revisiting the relationship between managerialownership and …rm performance

esti-In corporate …nance, a common empirical challenge is to disentangle the tionship of interest from other potential confounding e¤ects In particular, omittedvariables can bias classic econometric estimations and, more generally, cast doubt

rela-on any causal inferences being made by researchers These endogeneity crela-oncerns arepotentially severe within the context of my research and one of the main contribu-tions of my work is to devise two novel empirical approaches to circumvent, at leastpartially, the issue of endogeneity

In trade-o¤ theories of capital structure, it is argued that the tax advantages ofdebt are counter-balanced by costs of …nancial distress caused by a default of the …rm

on its debt obligations Although the costs of …nancial distress are a critical element

of our understanding of the use of debt by corporations, the reverse causality channel,

…nancial distress, hinders most empirical attempts to measure these costs accurately(e.g Kalay, Rajeev, and Tashjian (2007)).

In my …rst essay, I analyze a comprehensive sample of defendant …rms that foundthemselves exposed to an unexpected wave of asbestos litigation in the wake of twoU.S Supreme Court decisions Since these legal liabilities are unrelated to currentoperations, …rms that are in …nancial distress due to their legal woes provide a naturalexperiment to study the impact of …nancial distress on a …rm’s operational perfor-mance When analyzing …rms su¤ering from this exogenous shock to their …nances, I

…nd little evidence of negative spillover e¤ects ("indirect" costs) of …nancial distress.That is, the competitive position of the distressed …rms is not adversely impacted

by their weakened …nancial situation Furthermore, I …nd empirical support for asigni…cant disciplinary e¤ect of …nancial distress as these …rms actively restructureand refocus on core operations.(Wruck (1990))

In my second and third essays, I focus on the relationship between managerialownership and …rm performance In the second essay, I analyze in a systematicmanner the impact of …rm size heterogeneity on the relationship between managerialownership and …rm performance using a large panel dataset of U.S …rms over theperiod 1988-2004

I reconcile some of the extant literature by showing that the relationship is tive to the …rm size characteristics of the sample being used (see Kole (1995) for anearlier attempt) In particular, I recover the classic hump-shaped relationship (e.g.Stulz (1988)) when focusing only on the largest …rms, while the relationship turnsnegative when the sample is comprised of smaller …rms

sensi-I o¤er three potential explanations for the negative relationship among smaller

…rms First, given that managerial ownership levels among small …rms are oftenhigh enough to insulate them from the market for corporate control (see Holderness(2009)), the negative relation for small …rms is consistent with entrenchment argu-ments The second explanation is derived from the life-cycle theories of managerialownership (see Helwege, Pirinsky, and Stulz (2007)) I …nd that for lower levels ofmanagerial ownership, the negative relationship is driven by older …rms that have onaverage less liquid stocks This …nding is consistent with …rms that do not performwell enough to create a liquid market for their stock, and hence have to keep highlevels of insider ownership in order to avoid a negative price impact that would re-sult from a reduction of their stake Lastly, the di¤erences found in the relationshipacross …rm size categories are also suggestive of endogeneity issues These issues areinvestigated in my third essay, which I summarize below

Principal-agent models predict that managerial ownership and …rm performanceare endogenously determined by exogenous changes in a …rm’s contracting environ-ment (e.g Holmstrom (1979)) Changes in the contracting environment are, however,only partially observed, and the standard statistical techniques used to address endo-geneity may be ine¤ective in this corporate setting (see Coles, Lemmon, and Meschke(2007)) The typical unobservable variable in this context is managerial productivity

By driving both …rm performance and managerial ownership, this omitted variablecan bias the results derived from the typical frequentist approach

In my third essay, together with my coauthor Phil Davies, we develop a noveleconometric approach to control for the in‡uence of time-varying unobserved vari-ables related to a …rm’s contracting environment Using a large panel dataset of

U.S …rms over the period 1988-2004, we …nd no evidence of a systematic relationbetween managerial ownership and performance Our study thus reaches conclusionssimilar to those of Harold Demsetz and his coauthors (Demsetz (1983), Demsetz andLehn (1985), Demsetz and Villalonga (2001)) who argue that costs to changing thedistribution of ownership cannot be large enough to insulate …rms subject to marketforces from suboptimal ownership structures over a long period of time Our key con-tribution stems from obtaining these results while explicitly modeling for the e¤ect

of omitted variables in the context of the relationship between managerial ownershipand …rm performance

CHAPTER 2

THRIVING IN THE MIDST OF FINANCIAL DISTRESS?

AN ANALYSIS OF FIRMS EXPOSED TO ASBESTOS

Asbestos litigation is one of the most important mass tort litigations in the U.S.(see Carroll, Hensler, Gross, Sloss, Schonlau, Abrahamse, and Ashwood (2004)) This

1 The lack of consensus stems mostly from widely di¤erent estimates of the indirect costs of distress Weiss and Wruck (1998) …nd massive value destruction in Eastern Airline’s bankruptcy and Pulvino (1999) …nds evidence of signi…cant …re sales in the airline industry Maksimovic and Phillips (1998) …nd no evidence of reduced e¢ ciency in plant productivity during bankruptcy and Andrade and Kaplan (1998) also …nd a neutral e¤ect on …rms in …nancial distress due to prior highly leveraged transactions In contrast, Kalay, Rajeev, and Tashjian (2007) …nd signi…cant improvement

in the operating performance of …rms once they …le; concluding that Chapter 11 might yield net gains However, their sample of …rms might have su¤ered from signi…cant indirect costs of …nancial

study focuses on an unexpected surge in litigation activity after the U.S SupremeCourt overturned class action settlements in Amchem (1997) and Ortiz (1999) Due tothe long latency of asbestos-related illnesses, plainti¤s seek purely monetary compen-sations for injuries caused by operations that ceased more than twenty years earlier.Hence, defendant …rms experience a non-debt liability shock that is unrelated to cur-rent operations.2 As such, this setting provides a unique opportunity to analyze thecosts and net e¤ects of …nancial distress on …rms’real activities.

Using accounting disclosure rules that require …rms to report any material pendinglegal proceedings, I …nd 270 public corporations exposed to asbestos litigation in theU.S in the aftermath of the aforementioned U.S Supreme Court rulings Exposure

to asbestos litigation varies considerably within the sample For each …rm, I …rstdetermine the direct …nancial impact of the litigation Furthermore, I gauge the level

of …nancial constraints imposed by the litigation by determining the availability anddemand for funds and any resulting …nancial restructuring similarly to Kaplan andZingales (1997).3 After an extensive analysis of …nancial statements, press releasesand corporate policies (as in DeAngelo and DeAngelo (1990)), I …nd that most …rms

in the sample have not experienced any material …nancial consequences from thelitigation

An investigation of the mitigating devices available to defendant …rms revealsseveral interesting …ndings First, beyond the traditional legal recourse of settlingand …ghting claims in court, …rms are often completely shielded by their product

2 In sharp contrast, tobacco litigation also a¤ects the defendant …rms’ product market as the tobacco industry still produces the litigious goods (see Dahiya and Yermack (2003)).

3 Furthermore, as my study focuses on …nancial distress as opposed to economic distress, I also control for any potential confounding e¤ects that could explain a signi…cant deterioration in oper- ating performance.

and premises liability insurance coverage As they only bear out-of-pocket costs onceliability insurance coverage is depleted, many …rms in the sample report little or nonet litigation liability Second, Chapter 11 is sometimes used preemptively by …rms

to settle their legal liabilities while shielding operations by …ling only their laden unit or by using an a¢ liated …rm’s Chapter 11 proceedings These …rms thentry to extend the bankruptcy court protection to all subsidiary and parent companies

asbestos-in order to obtaasbestos-in the automatic stay asbestos-in litigation that stops the cash ‡ow draasbestos-in Iuncover a dozen companies that make use of Chapter 11 while not su¤ering from anyvisible curtailments in the access to their usual sources of credit.4

Fifteen …rms do become …nancially distressed due to their exposure to asbestoslitigation While these defendant …rms have positive operating margins during thedistress period, their rapidly increasing legal liabilities and cash out‡ows combinedwith the general uncertainty surrounding the ultimate cost of asbestos legacies lead

to a severe weakening of their …nancial position I …rst estimate the net costs of

…nancial distress among this sample of …rms The direct costs of …nancial distress(e.g professional fees, etc.) average around 8% of pre-distress …rm value Thenet indirect costs of …nancial distress for the distressed sample are estimated usingvalue-based methods as in Andrade and Kaplan (1998) Industry-adjusted measuresshow a statistically insigni…cant median gain of …nancial distress of about 15% to30% of pre-distress …rm value depending on the discount factor used Overall, …rmvalue is preserved This result seems at odds with well established corporate …nancetheories For instance, …nancial distress is thought to a¤ect operating performance by

4 A major advantage of a bankruptcy …ling is to o¤er the only de…nite release from present and future asbestos liability through what is called the “channeling injunction”, whereby a bankrupt

…rm sets up and …nances a Trust to compensate claimants and emerge from bankruptcy free of any

worsening the competitive position of the …rm and forcing it to cut back on pro…tableprojects (Bolton and Sharfstein (1990)) Furthermore, …nancial distress could have

a disruptive e¤ect on the relationships the …rm has with its suppliers, key employeesand customers (see Parsons and Titman (2008) for a review) So to understand thisresult, I take advantage of the small sample size to perform a holistic analysis of theimpact of …nancial distress on the …rms’ operations (e.g DeAngelo and DeAngelo(1991))

First, I investigate how these …rms manage to reestablish their sources of …nancing.Banks become the lenders of last resort during the crisis Three …rms are able torenegotiate their debt in out-of-court …nancial restructurings In the 12 other cases,Chapter 11 is used to obtain an easier access to new capital by giving new creditorsseniority over the existing ones Strikingly, all Chapter 11 …lings of entire businessesobtain a Debtor-In-Possession (DIP) …nancing (see Dahiya, Kose, Puri, and Ramirez(2003)) and all partial …lings (of tainted subsidiary) obtain a renewed credit facility.After a decline at the onset of distress, operating performance recovers over thecrisis period At the resolution of distress, operating performance is on average back

to its pre-distress levels or higher, both on a nominal and industry-adjusted basis It

is explained in part by better productivity and cost cutting There is an economicallyand statistically signi…cant drop in the number of employees per sales over the period

of about 25% from pre-distress levels while market share drops only slightly during thedistress period Looking at patterns in discretionary spending, I …nd a signi…cant dropafter the onset of distress in both capital expenditures and M&A activity followed by

a mild recovery in the case of capital expenditures

These quantitative results are interpreted in light of qualitative evidence Withrespect to costs of …nancial distress, there is some evidence of costly CAPEX cur-tailments but only in a minority of cases and they tend to be temporary, occurringaround the onset of distress or shortly after Evidence of …re sales is limited There

is no evidence of risk-shifting behavior among the distressed …rms In contrast, I …ndwidespread evidence of potential bene…ts of …nancial distress through its disciplinarye¤ects on …rms (Wruck (1990)) Almost all …rms take the opportunity to shed non-core assets, restructure current operations, often with strong operationally-orientedmanagement at the helm; replacing previous management at the onset of distress inabout half the cases Firms strive to keep new management in place by o¤ering KeyEmployee Retention Plans or “KERP” with cash based long-term incentives based

on operational performance

One potential explanation for these observed operational improvements relies onthe fact that an increase in bank monitoring combined with a surge in legal liabilitiesand related cash out‡ows can act as triggers for management to restructure andavoid any wasteful projects (Jensen (1986)) Because these …rms have the capacity togenerate high levels of free cash ‡ow and show prior signs of weak internal governancemechanisms, the e¤ect of increased legal liabilities could be viewed similarly to that

of the disciplinary e¤ects caused by a leveraged management buyout or a leveragedrecapitalization (see Kaplan (1989) and Wruck (1994)) My empirical results also lendsupport to Haugen and Senbet (1978)’s conjecture that only assets of poor qualityand the associated risk of liquidation lead to indirect costs of …nancial distress andthat claimants can reorganize a …nancially distressed …rm without destroying …rmvalue, as long as stakeholders rationally anticipate the transfer of ownership

This paper contributes to the literature estimating the costs of …nancial distress.Many papers analyze the costs of …nancial distress (e.g.Altman (1984), Asquith,Gertner, and Scharfstein (1994), Opler and Titman (1994), Maksimovic and Phillips(1998)) However, their estimates of the net indirect e¤ects of …nancial distress areclouded by the fact that these studies cannot distinguish …nancial distress from eco-nomic distress For instance, Kalay, Rajeev, and Tashjian (2007) document operatingimprovements during Chapter 11 However, the leverage of these …rms is signi…cantlyhigher than their comparison group already seven years prior to …ling and the de-terioration in operating performance steadily declines over the …ve years prior tobankruptcy Arguably, part of the costs of …nancial distress have been incurred prior

to the bankruptcy …ling date, yet they cannot be measured precisely in their settingbecause, as stated in Wruck (1990), “it is impossible to tell whether the loss in pro…ts

is in fact caused by …nancial distress or whether …nancial distress is caused by theloss in pro…ts.”

Several papers have tried to identify pure …nancial events For instance, Lamont(1997) and more recently Rauh (2006) …nd that …rms’ investment policies can besensitive to an exogenous shock to cash ‡ows, thus highlighting one potential indirectcost of …nancing constraints on real activities Using an intra-…rm litigation setting,Cutler and Summers (1988) and Bhagat, Bhagat, Brickley, and Coles (1994) hypothe-size that a combined market value loss in excess of the litigation costs re‡ects indirectcosts of …nancial distress However, they do not provide supportive evidence at the

…rms’operational level

The paper most related to this study is provided by Andrade and Kaplan (1998)who use Highly Leveraged Transactions (HLTs) to detect …nancial distress, as opposed

to economic distress However, in their study, …rms choose a highly leveraged capitalstructure prior to the distress Beyond this self-selection issue, the HLTs triggerorganizational changes and strong incentive e¤ects of their own (e.g Jensen (1989))and as such provide another confounding e¤ect when trying to measure the impact of

…nancial distress Given the unique empirical setting of my study, I provide furtheruseful evidence of the potential bene…ts and costs of …nancial distress.5

The rest of the paper is organized as follows Section 1 describes the institutionalbackground of the asbestos crisis Section 2 presents the data and Section 3 estimatesthe cost of litigation for defendant …rms while analyzing the mitigating devices used

to lessen the …nancial impact the litigation Section 4 estimates the costs of …nancialdistress and gives a detailed analysis of the performance and behavior of …rms in

…nancial distress due to the litigation Section 5 concludes

2.2 Asbestos mass tort litigation

Asbestos mass tort litigation is characterized by the following two features First,there is a long latency period between exposure to asbestos and the potential manifes-tation of one of its related illnesses As such, asbestos litigation is usually unrelated

to current underlying operations of defendant …rms and thus produces an exogenousshock to their …nances Second, due to its impressive physical properties and low cost,asbestos was used in a wide range of applications ranging from insulation products tobrakes in automobiles Its use in manufacturing processes was drastically curtailed

5 A contemporaneous study also seizes on the opportunity o¤ered by asbestos litigation to study

an exogenous shock to …rms’ …nancial strength (Hadlock and Sonti (2009)) They …nd that the stock value of competitors falls at the announcement of defendant …rms’large legal liabilities This evidence could be interpreted as a reassessment by the market of an increased probability that …rms within the same industry will also be taken into the sprawling litigation It could also be evidence that increased …xed liabilities lead to more aggressive competitive interactions, which would be

only as of the mid-’80s due to the health risk and associated legal liability (White(2004)) These facts explain why the pool of defendant …rms comes from a largecross-section of di¤erent industries, extending well beyond asbestos producers Takentogether, these characteristics justify the use of this litigation setting to analyze theconsequences of a negative exogenous shock to …rms’…nancial strength.

2.2.1 The asbestos litigation crisis after Amchem (1997) and

Ortiz (1999)

Asbestos is the longest-running mass tort litigation in the United States, spanningover more than four decades.6 As of the …rst half of the 1970s, courts ruled thatplainti¤s could …le civil suits against their past employers, which had been protected

by workers’ compensation until then Furthermore, with “strict liability” verdicts,manufacturers become liable as long as causation can be proven And causationcan be argued more convincingly in the case of asbestos as exposure leads to so-called “signature” diseases such as mesothelioma This fact partly explains the rise

in litigation activity aiming at …rms with a more tenuous link to past asbestos useonce most asbestos producers had gone bankrupt.7

Furthermore, with “joint and several” liability, a plainti¤ may recover all thedamages from any of the defendants regardless of their individual share of the liability.Hence, by the early ’90s, plainti¤s’ attorneys had clearly de…ned legal strategies inplace: they bundled cases together and …led claims in plainti¤-friendly state courts

6 Refer to the extensive RAND study (Carroll, Hensler, Gross, Sloss, Schonlau, Abrahamse, and Ashwood (2004)) and the references therein A detailed timeline of asbestos litigation in the U.S can be obtained from the author.

7 The …rst waves of bankruptcies concerned mostly asbestos producers Since the litigation fected both the …rms’ …nances and their operations, they are not part of our study For example, Johns Manville …led for bankruptcy in August 1982 but stopped its asbestos-producing operations only in 1983, selling the unit at a loss in 1985 while still in bankruptcy.

af-against many …rms at a time (see White (2002a)) By that time, asbestos personalinjury claims accounted for up to a third of all civil …lings for some courts (seeHensler (2002)) Due to the potentially crippling e¤ect of losing against thousands

of claimants in trial, most defendant …rms aimed at striking deals with large pools

of (mostly unimpaired) plainti¤s, usually in out-of-court settlements Meanwhile, noglobal government-sponsored solution had emerged Two e¤orts were made by majordefendant …rms to …nd an e¢ cient solution using class action status.8 However, theseclass action settlements were contested on legal grounds and were rejected by the U.S.Supreme Court in Amchem Products v Windsor in 1997 and Ortiz v FibreboardCorp in 1999

In the wake of the U.S Supreme Court rulings, all claims were reinstated in thecourts Combined with an unexpected surge in the number of new lawsuits sparked

by heightened media attention, several key defendant …rms saw their asbestos liabilitysoar The only viable option left for some of these …rms was to seek protection frombankruptcy courts with Owens Corning’s …ling being the most notable in October

2000 However, the stay in litigation o¤ered by the bankruptcies e¤ectively tookmoney for compensation o¤ the table, thus plainti¤ attorneys aggressively sought outnew defendants These events led to an unprecedented level of litigation activity (seeWhite (2002b)) Sometimes called the “fourth wave” in asbestos litigation, it is theevent I study in this paper.9

8 In 1993, the Center for Claims Resolution (CCR), set up by 20 co-defendants, settled with most of its current plainti¤s and provided a settlement for all future claims through a class action settlement (known as the “Georgine Settlement”) Fibreboard, another traditional defendant, proposed a class action settlement of its own the same year.

9 Benston (2003) Benston (2003) and Stiglitz, Orszag, and Orszag (2002)o¤er an analysis of …rms hit by asbestos litigation as the crisis was unfolding To date there is still no global solution set forth

by Congress However, a slew of state reforms have led to an abatement of the severity of asbestos

2.3 Data

The initial sample selection aims at detecting the broadest cross-section of …rmssu¤ering from exposure to asbestos litigation in the wake of the de…ning U.S SupremeCourt decisions in Amchem (1997) and Ortiz (1999) Accounting disclosure rulesrequire …rms to report any material pending legal proceedings I perform a key wordsearch regarding exposure to U.S asbestos litigation within …nancial statements ofU.S and foreign publicly listed …rms over the period from 1996 to 2005.10

From this initial sample, I eliminate …rms that only report taking a charge linked

to the abatement of asbestos in their buildings, including real estate …rms I alsodiscard certain industries, namely: utilities, which are inappropriate due to their reg-ulated nature, asbestos abatement …rms due to the potential impact of litigation ontheir core operations and tobacco …rms due to their own litigious products Financialservices …rms and in particular insurance carriers and reinsurers that covered asbestosliabilities have incurred large cash out‡ows linked to the litigation However, provid-ing coverage is part of their core operations and as such, these …rms do not …t theresearch design and are excluded from the sample The search produces 270 …rms

10 In more detail: I used the 10-K WizardR SEC Power Search for a key word search among all 10-K and 20-F SEC …lings and hand-checked every …rm that was ‡agged by at least one of the following queries

(*: denotes stemming):

1 The word “asbestos” within 500 words of the word “litigation”

2 The word “asbestos” within 20 words of the word “liabilit*”

3 The word “asbestos” in item 3 of part I of 10-Ks (i.e in Legal Proceedings)

4 The word “asbestos” in the Management’s Discussion Section.

5 The word “asbestos” within 500 words of “commitment*” or “contingenc*” in item 8 of Part

II (i.e in Financial Statements and Supplementary Data)

For foreign …rms, the search is restricted to those that have exposure to U.S litigation where the timeline and legal framework are well understood and common to the other …rms in the sample The sample was partially cross-referenced using a comprehensive list of all bankrupt …rms where asbestos was mentioned as a reason for Chapter 11 …lings in Carroll, Hensler, Gross, Sloss, Schonlau, Abrahamse, and Ashwood (2004) It was further compared to all …rms mentioned at least once in the WSJ articles containing the word “asbestos” during the period of study.

citing asbestos litigation concerns during the period under study.11 Panel A of Table2.1 provides summary characteristics of these defendant …rms relative to the universe

of …rms from the CRSP/Compustat dataset that do not report being sued for pastasbestos use Defendant …rms are more mature than their counterparts This is ex-empli…ed by the fact that they are larger, have more tangible assets, pay out more andhave slower growth than non-defendant …rms The fact that only …rms that used as-bestos in the past can be successfully sued explains why this is the case Furthermore,defendant …rms are slightly more leveraged but have higher interest coverage levelsbecause they are globally more pro…table Taken together, these …rm characteristicshighlight what is known as the “deep pocket” syndrome, whereby plainti¤ attorneystend to sue …rms with the ability to pay large settlements or damage awards

Panel B of Table 2.1 gives the evolution of the number of …rms reporting exposure

to asbestos litigation and the corresponding number of a¤ected industries measured

at the Standard Industry Classi…cation (SIC) two-digit level Figure 2.1, based onthese numbers, highlights the sprawling nature of the litigation after 1999 with a neartripling of …rms reporting being defendants Panel B of Table 2.1 also reports thethree industries most hit by the litigation; they represent on average about a third

of all cases Not surprisingly, building materials, steel work and chemicals are oftenamong the list as these industries were heavy users of asbestos in the past

11 The small sample size relative to the overall number of defendants cited by Carroll, Hensler, Gross, Sloss, Schonlau, Abrahamse, and Ashwood (2004) is due to my focus on (1) the most recent wave of the litigation and (2) …rms with publicly traded debt or equity, which do not need to report their exposure if immaterial.

2.4 Financial impact of litigation on defendant …rms

In this section, I analyze defendant …rms and categorize them by their …nancialexposure to the litigation I quantify the …nancial impact of the litigation by gaugingboth the direct legal costs and the indirect impact on defendant …rm’s access tosources of funding In doing so, I highlight several mitigating tools used by defendant

…rms to lessen the …nancial burden caused by the litigation

2.4.1 Measuring the …nancial impact of litigation beyond

in-surance coverage

I rely on …nancial statements and general disclosure regarding asbestos litigation

to construct both qualitative and quantitative measures of …nancial exposure to thelitigation Legal liabilities can be viewed as non-debt liabilities owned by a dispersedgroup of claimants (the plainti¤s) The qualitative evidence stems from managementreporting on the issue The quantitative evidence is found in …nancial statementswhen the litigation leads to signi…cant expenditures, liabilities and …nancial restruc-turings, including equity dilution in Chapter 11 reorganizations.12

Figure 2.2 highlights the multiple mitigating devices defendant …rms can use tolimit their …nancial exposure to the litigation As the …gure illustrates, the …rst line

of defense for defendant …rms is provided by their insurance policies or self-insuranceprovisions Asbestos coverage, as part of Comprehensive General Liability (CGL)policy, covers product and premises liability incurred up to an aggregate limit.13

12 SFAS 5 requires pending claims that still have to be settled or tried to be accrued as an expense

if certain conditions are met The conditions being: (1) if it is probable that a liability has been incurred and (2) the amount of such liability incurrence can be reasonably estimated.

13 As a consequence of the CGL crisis of 1985-1986, the “absolute” pollution exclusion clause was added to the standard policies, which led to the exclusion of asbestos coverage after that date (Abraham (2001)) Because a majority of claims have exposures to asbestos prior to 1986, most …rms

These policies also extend to pay for defense costs with insurers usually deciding onthe optimal litigation strategy to adopt Furthermore, …rms often contract multiple

“layers”of insurance Once the primary coverage is exhausted, …rms may access theirupper layers, called “excess”layers for further coverage Legal liabilities signi…cantlyimpact the …rms’ …nances only once the coverage is exhausted Hence, I computemeasures of exposures both on a gross and net of insurance coverage basis

Speci…cally, I construct an annual time-series of both a ‡ow and a stock measure

of exposure to the litigation at the …rm level by computing annual (1) gross and netcash out‡ows, i.e cash out‡ows to the plainti¤s beyond insurance recoveries in thelatter case and (2) gross and net legal liabilities, i.e liabilities related to the litigation,minus the insurance receivables from the …rm’s insurance policies in the latter case

I also infer an annual “perfect foresight” measure of gross and net liability whichcomputes the sum of all gross and net expenditures linked to the litigation from eachyear up to the last year of the sample, to which I add any remaining book gross ornet liability stated at the end of the sample period in …scal year 2006.14 This measuregives the liability with perfect foresight, which might di¤er from a …rm’s estimatedliability at the time

Table 2.2 provides the categorization of …nancial exposure within the sample cording to both the sum of net out‡ows (dollar value of shares transferred to plainti¤splus net cash out‡ows) relative to the sum of operating pro…ts (EBITDA) over the

ac-in the sample are still able to get coverage from their ac-insurance policies Had ac-insurance coverage been available after 1986, it can be assumed that policy premiums would have become prohibitively expensive as insurers would have adapted them in the face of mounting litigation activity.

14 I expand the time period of analysis relative to the key word search period in order to have a more complete picture of the …nancial exposure Fiscal 2006 is the resolution year for most of the largest cases The classi…cation of exposure does not change if a 5% rate is used to discount these forward-looking ‡ows and liabilities.

period 1999-2006 (‡ow measure) and the maximum “perfect foresight” net legal bilities relative to total assets during the period 1999-2006 (stock measure) From theinitial 270 …rms, a majority of …rms have only immaterial exposure to the litigation.Approximately two thirds of the …rms citing exposure to asbestos litigation do notrecord any net liabilities or out‡ows (category 1 …rms in Table 2.2) Most of themstate having insurance coverage in case the litigation were to become more important

lia-in the future Disclosure from these “peripheral”defendants tends to be limited; ically they state being named among dozens, sometimes hundreds of co-defendantswithout having any of their products being speci…cally blamed Caterpillar Inc issuch a …rm From information disclosed in a court ruling (see Bronte (2007)), Cater-pillar has expended a total of about $25 million on asbestos litigation over the periodranging from the late ’80s to 2004 This amount can be considered as immaterial for

typ-a corportyp-ation of its size typ-and expltyp-ains why it htyp-as never disclosed typ-any litigtyp-ation-reltyp-atednumbers in its …nancial statements

In all, 82 …rms report either a liability and/or cash out‡ows beyond insurancecoverage or self-insured provisions made in relation to the litigation over the period1999-2006 (category 3 to 10) No …rm in the sample su¤ers from …nancial distresswithout having exhausted, at least partially, its insurance coverage Insurers’insol-vency, in particular among the providers of excess insurance layers, is reported as acause for loss of full coverage in 20% of the cases Among the insured …rms reporting

a net out‡ow or liability, about 65% of them mention being in litigation with theirinsurance carriers to determine the extent of coverage under the policies contracted(see Anderson (1987) and McGuire, McCullough, and Flanigan (2004))

Among the defendant …rms reporting out-of-pocket costs, more than a third ofthem report a net liability but no cash out‡ows over the period (category 3 of the ‡owmeasure) Most …rms in this category have immaterial annual out‡ows but establish

a net liability to re‡ect their legal obligations in the years to come Furthermore,among …rms reporting out‡ows, a great majority of them pay out less than 5% oftheir accumulated operating pro…ts over the period (categories 4 and 5) Hencethe distribution of …nancial exposure among defendants is heavily skewed with mosthaving only minimal exposure Less than two dozens key defendants have to bearmaterial legal liabilities and su¤er net out‡ows due to their past asbestos use (category

6 and beyond for both the stock and ‡ow measures)

2.4.2 Litigation exposure and …nancial distress

To determine if the litigation exposure caused distress among the sample …rms,

I rely on a detailed analysis of the …nancing and liquidity situation for each sample

…rm as in Kaplan and Zingales (1997).15 A …nancially distressed defendant …rm isde…ned as: (1) having a …nancial restructuring inside or outside of Chapter 11 and (2)showing evidence of being cut out of its usual sources of credit, due to asbestos litiga-tion and the uncertainty surrounding the ultimate legal costs The assessment relies

on disclosure from press reports, Moody’s credit reports and …nancial statements (inparticular Management’s discussion on liquidity and …nancing), on their access to dif-ferent public …nancial markets and private debt (lines of credit, covenant violations)and ancillary information contained in variables such as dividend and share repur-chase policies (see DeAngelo and DeAngelo (1990)) As this study focuses on …nancial

15 Typical criteria of …nancial distress such as a …rm’s inability to meet its interest requirements (e.g Andrade and Kaplan (1998)) are not directly applicable because the …nancial distress studied

distress as opposed to economic distress, I also document any potential confoundinge¤ects that could lead to operating di¢ culties within the sample of …rms In partic-ular, …rms with signi…cant asbestos litigation exposure that have at least one year ofnegative operating cash ‡ows (proxied by EBITDA adjusted for any asbestos-relatedprovisions or gains) during the period of analysis are also considered in economic dis-tress and the analysis is performed with and without them for robustness purposes Ifthe corporation …les for Chapter 11, asbestos litigation needs to be listed as a majorcontributor for the …ling.16

In the context of mass tort litigation, Chapter 11 may be used preemptively and isnot necessarily synonymous with …nancial distress (see Resnick (2000)) As Figure 2.2highlights, a …rm will try to …le either the asbestos-tainted subsidiary, use an a¢ liated

…rm’s Chapter 11 (e.g past owner of the business) or …le only the parent holdingcompany for bankruptcy before …ling the entire …rm into Chapter 11 The corporationwill then try to have the bankruptcy protection extend to all subsidiary and parentcompanies However, these strategic …lings are not always successful: Bankruptcycourts can “pierce the corporate veil” and hold the entire …rm liable for the legalliabilities of the tainted subsidiary There are no set rules in the matter and thejudge overseeing the proceedings has broad authority over the matter.17 If successful,plainti¤s stop being paid as the …rm obtains an automatic stay in litigation The

16 Only two …rms have material asbestos legal liabilities and severe operating di¢ culties unrelated

to the litigation Several …nancially distressed …rms, such as Foster Wheeler and Bethlehem Steel, are commonly associated with asbestos litigation yet their operational turmoil dwarfs their asbestos exposure These …rms are not included in the sample of …nancially distressed …rms due to asbestos litigation A detailed list of these …rms can be obtained by contacting the author.

17 For instance, a case ruling in the ABB bankruptcy reduced the scope of the bankruptcy tion that was previously extended to non-bankrupt subsidiaries As a result, Crane put o¤ its plans

protec-to …le for bankruptcy MCC, its asbesprotec-tos-laden unit In another example, the Big Three auprotec-tomakers were not allowed to settle their claims within the Federal Mogul Chapter 11 proceedings.

automatic stay stops the cash ‡ow drain In any case, the long-term advantage of thebankruptcy …ling is to o¤er the only de…nite release from asbestos liability throughwhat is called the “channeling injunction”.18

The “channeling injunction”allows a …rm to emerge from Chapter 11 bankruptcydischarged of all present and future liability by setting up a Trust, which needs to

be approved by 75% of the asbestos creditor class Once the Plan is accepted ande¤ective, the debtor receives a "channeling injunction" that bars all claims againstthe reorganized …rm and redirects them to the Trust The Trust serves to compensateall the present and future claimants and is usually funded by a combination of theemerged …rm’s stock, cash, notes and remaining insurance policies.19

Table 2.3 provides the results of the detailed …nancial analysis for the 30 mostexposed …rms listed according to the “perfect foresight”net liability measure of Table2.2.20 Because …rms hit by the litigation tend to be large and pro…table with manysources of funding, only the ones submerged by litigation liabilities (e.g USG, OwensCorning) and those having high prior leverage (e.g ABB, GP) su¤er from …nancialdistress unrelated to their current business operations Strikingly, only 15 out of the 30

…rms with the most signi…cant asbestos liabilities report severe …nancing constraintsdue to asbestos litigation (in bold in Table 2.3) The other …rms display several signsthat show they are not constrained in their …nancing during the period of analysis Forinstance, Crane, RPM International, Dow Chemical or Honeywell increase dividends

18 Also known as the Manville amendment, Congress rendered the “channeling injunction”o¢ cial through amendments in the Bankruptcy Reform Act of 1994 (see U.S.C § 524(g)).

19 The Trust has to receive the equivalent of 50.1% of the value of the pool of assets covering the liability Although the Bankruptcy judge has discretion in that regard, existing shareholders risk a signi…cant dilution of their equity stake in the proceedings.

20 None of the …rms reporting a smaller “perfect foresight” net liability su¤ered from …nancial distress due to the litigation.

throughout the period, do not draw down on their lines of credit and/or use publicdebt markets on a regular basis.

Furthermore, several …rms with signi…cant asbestos liabilities did not su¤er fromany …nancial distress due to their astute use of the Chapter 11 forum.21 The case

of PPG Industries is illustrative PPG …led for Chapter 11 its co-owned subsidiaryPittsburgh Corning (PC) in April 2000, soon after the …rst case where it was foundliable in January 2000 Although the Trust settlement negotiated within the Chapter

11 proceedings leads the …rm to report net liabilities of close to $900 million as of

…scal year 2006, its out-of-pocket costs have been immaterial to date Furthermore,the …rm does not su¤er from any …nancing constraints over the period of analysis: Ithas signi…cant untapped credit lines and generates su¢ cient cash ‡ow from operations

to maintain an S&P credit rating of A, increase dividends per share and retire moredebt than it issues throughout the period.22

Enpro Inc is the only successful spin-o¤ of an asbestos-tainted subsidiary amongasbestos defendants.23 This type of divisional restructuring is of limited use for thefollowing reasons First, asbestos liabilities need to be clearly contained within one ofthe business units in order to shield the rest of the operations from further litigation.Second, even if the …rm can isolate the tainted unit, it needs to …nd a buyer who iswilling to take on the liability; no small feat if the liability is known, especially given

21 However, some of these …rms do su¤er from credit downgrades for other reasons For example, Corning Inc faced restrictions in its …nancing and downgrades due to the massive downturn among telecommunication equipment providers amid the bursting of the technology bubble.

22 Details on all the …rms that use Chapter 11 strategically and that have not su¤ered any …nancial distress as a consequence of the litigation can be obtained by contacting the author.

23 It was spun-o¤ by Goodrich and although it had signi…cant asbestos liabilities, its parent deemed an outstanding debt issue prior to the spin-o¤ in order to minimize the risk of bankruptcy.

re-the risk of successor liability.24 Lastly, even when the …rm can merge the tainted unit

or otherwise spin it o¤ to existing shareholders, it can still be sued for fraudulentconveyance if the subsidiary subsequently defaults

Gormley and Matsa (2009) …nd that …rms with weaker balance sheets exposed to

a product liability shock try to “grow out” of their legal liabilities, often by makingdiversifying acquisitions This mitigating strategy seems at odds with the fact thatplainti¤s’ attorneys tend to sue “deep pockets” and I …nd no such evidence in mysample The dichotomy can be explained by the fact that in the wake of the U.S.Supreme Court decisions, plainti¤s’attorneys sued indiscriminately; even …rms withthe most remote link to asbestos became targets As such, asbestos defendant …rmsdid not have the same warning to diversify or grow away from their legal liabilities

as the carcinogen producers did in their study

To conclude this section, two striking features are worth underlining First, many

…rms have immaterial …nancial exposure to asbestos litigation It can be explained bythe fact that most defendants have insurance coverage for asbestos liability and tend

to be peripheral defendants with only a tenuous link to past asbestos use Second,among the subset of 82 …rms reporting net legal costs, a dozen manage to use Chapter

11 preemptively and do not report constrained access to external funds speci…callydue to the litigation As a consequence of these two facts, I uncover only 15 …rmsthat su¤er from …nancial distress due to asbestos litigation These 15 …rms are thefocus of the analysis from this point on

24 For instance, WR Grace had agreed to retain the asbestos-liability when it merged its packaging unit with Sealed Air but later …led for Chapter 11 under the weight of its asbestos legacy The Bankruptcy court allowed plainti¤s to sue Sealed Air for fraudulent conveyance, which led it to settle for more than $800M.

2.5 Analysis of …nancially distressed …rms due to asbestos

litigation

In this section, I perform a thorough analysis of the sample of …rms su¤ering from

…nancial distress due to their asbestos liabilities First, I give a description of thesample, in particular with regards to their source of exposure, their …rm characteris-tics, how the liability is resolved and the timeline in litigation Second, I analyze thecost of litigation and the direct cost of …nancial distress Third, value computationsare performed to infer net indirect costs of …nancial distress Finally, these valuecomputations are put in the context of the …rms’ operating and …nancing behaviorduring the period of distress using both qualitative and quantitative evidence

2.5.1 Characteristics of …nancially distressed …rms

Table 2.4 provides a description of the sample of 15 …rms su¤ering from …nancialdistress due to asbestos litigation The table is divided into two panels Panel Alists the 12 …rms using Chapter 11 and Panel B lists the three …rms that rely on anout-of-court …nancial restructuring to resolve their …nancial crisis

Various observations can be made from the table First, a third of the distressed

…rms acquire most of their litigation exposure during the second part the ’90s, even

as late as 1999 The fact that by then asbestos litigation had been ongoing for morethan three decades lends support to the unanticipated nature of the most recent wave

in litigation under study

Second, the industries hit by the litigation can be considered as quite diverse,ranging from oil and gas to power and automation This fact is a re‡ection of plainti¤attorneys aggressively pursuing more peripheral defendants after the main producers

of asbestos had gone bankrupt Past heavy users of asbestos such as building materialsand refractory …rms became the next target, which explains why …ve out of the 15

…rms operate in building materials All …rms are sued for their past use of asbestoseither in the workplace or in their products For instance, Crown, Cork and Seal,

a major packaging …rm, is being sued for having owned for three months a smallinsulating …rm which used some asbestos in its products during the ’60s.25

Third, there is further evidence among this subset of distressed …rms of the use

of Chapter 11 to shield business divisions not tainted by asbestos when possible.Out of the …ve cases where the exposure to the litigation is concentrated within onespeci…c business, four …le only the asbestos-laden unit for Chapter 11, shielding tosome degree the rest of the …rm’s assets.26 In the case of BMCA, its parent G-IHoldings …les for Chapter 11 allowing BMCA to obtain the stay in litigation eventhough it remains outside of Chapter 11 itself.27 However, in 10 out of the 15 cases,the …rms are sued for successor liability and no speci…c current business operationscan be pinpointed as being the asbestos-laden unit

In Table 2.5, I compare …rms in …nancial distress due to the litigation with thenon-…nancially distressed among the sample of 82 …rms reporting a net out‡ow orliability over the period 1999-2006 (category 3 to 10 of Table 2.2)

25 A detailed account of the source of exposure to the litigation for the …nal sample of …rms can

be obtained from the author.

26 RHI deconsolidated its U.S operations and …led them for Chapter 11, in e¤ect shielding its healthy non-U.S based refractory operations while seriously weakening its overall …nancial strength.

27 BMCA is the only private …rm in the sample, …ling …nancial disclosure due to the public debt

it issues G-I holdings owns 100% of BMCA’s equity BMCA agreed to pay the …rst $204M in the litigation In exchange, G-I holdings agreed to indemnify BMCA for any excess liability When its parent …led, BMCA lost another $106M in receivables, was heavily downgraded by the rating agencies and had to renegotiate its credit facility It is still unclear who will become the owner of BMCA upon the emergence of G-I holdings.